Qualcomm® AI HubAI Hub

Profile Job Results

Jobs
jgnwxmyqg
Results Ready
Name
whisper_large_v3_turbo_quantized_decoder
Target Device
  • Dragonwing RB3 Gen 2 Vision Kit
  • Qualcomm Linux 1.6
  • Qualcomm® Dragonwing™ RB3 Gen 2 | QCS6490
Creator
ai-hub-support@qti.qualcomm.com
Input Specs
input_ids: int32[1, 1]
attention_mask: uint16[1, 1, 1, 200]
k_cache_self_0_in: uint8[20, 1, 64, 199]
v_cache_self_0_in: uint8[20, 1, 199, 64]
k_cache_self_1_in: uint8[20, 1, 64, 199]
v_cache_self_1_in: uint8[20, 1, 199, 64]
k_cache_self_2_in: uint8[20, 1, 64, 199]
v_cache_self_2_in: uint8[20, 1, 199, 64]
k_cache_self_3_in: uint8[20, 1, 64, 199]
v_cache_self_3_in: uint8[20, 1, 199, 64]
k_cache_cross_0: uint8[20, 1, 64, 1500]
v_cache_cross_0: uint8[20, 1, 1500, 64]
k_cache_cross_1: uint8[20, 1, 64, 1500]
v_cache_cross_1: uint8[20, 1, 1500, 64]
k_cache_cross_2: uint8[20, 1, 64, 1500]
v_cache_cross_2: uint8[20, 1, 1500, 64]
k_cache_cross_3: uint8[20, 1, 64, 1500]
v_cache_cross_3: uint8[20, 1, 1500, 64]
position_ids: int32[1]
Completion Time
11/14/2026, 12:39:20 PM
Options
--qairt_version latest
Versions
  • QAIRT: v2.51.0.260929181845
  • QNN Backend API: 5.51.0
  • QNN Core API: 2.40.0
  • Qualcomm Linux: 1.6-ver.1.2
  • AI Hub: aihub-2026.11.06.1
Estimated Inference Time
18.7 ms
Estimated Peak Memory Usage
15 ‑ 36 MB
Compute Units
NPU
1533
StageTimeMemory
First App Load
415 ms4‑5 MB
Subsequent App Load
368 ms2‑3 MB
Inference
18.7 ms15‑36 MB
QNNValue
context_options.htp_options.performance_modeBURST
default_graph_options.htp_options.optimizations[0].typeFINALIZE_OPTIMIZATION_FLAG
default_graph_options.htp_options.optimizations[0].value3.0
default_graph_options.htp_options.precisionFLOAT16
default_graph_options.htp_options.vtcm_size0

Sign up to run this model on a hosted Qualcomm® device!

Run on device